|
50
|
1
|
|
|
2 /*
|
|
|
3 * Copyright (C) Igor Sysoev
|
|
|
4 */
|
|
|
5
|
|
|
6
|
|
|
7 #include <ngx_config.h>
|
|
|
8 #include <ngx_core.h>
|
|
|
9 #include <ngx_http.h>
|
|
|
10
|
|
|
11
|
|
206
|
12 #define NGX_HTTP_NO_CHARSET -2
|
|
244
|
13 #define NGX_HTTP_CHARSET_VAR 0x10000
|
|
206
|
14
|
|
|
15 /* 1 byte length and up to 3 bytes for the UTF-8 encoding of the UCS-2 */
|
|
|
16 #define NGX_UTF_LEN 4
|
|
|
17
|
|
|
18 #define NGX_HTML_ENTITY_LEN (sizeof("") - 1)
|
|
78
|
19
|
|
|
20
|
|
50
|
21 typedef struct {
|
|
206
|
22 u_char **tables;
|
|
|
23 ngx_str_t name;
|
|
72
|
24
|
|
206
|
25 unsigned length:16;
|
|
|
26 unsigned utf8:1;
|
|
50
|
27 } ngx_http_charset_t;
|
|
|
28
|
|
|
29
|
|
|
30 typedef struct {
|
|
206
|
31 ngx_int_t src;
|
|
|
32 ngx_int_t dst;
|
|
78
|
33 } ngx_http_charset_recode_t;
|
|
|
34
|
|
|
35
|
|
|
36 typedef struct {
|
|
206
|
37 ngx_int_t src;
|
|
|
38 ngx_int_t dst;
|
|
|
39 u_char *src2dst;
|
|
|
40 u_char *dst2src;
|
|
50
|
41 } ngx_http_charset_tables_t;
|
|
|
42
|
|
|
43
|
|
|
44 typedef struct {
|
|
206
|
45 ngx_array_t charsets; /* ngx_http_charset_t */
|
|
|
46 ngx_array_t tables; /* ngx_http_charset_tables_t */
|
|
|
47 ngx_array_t recodes; /* ngx_http_charset_recode_t */
|
|
50
|
48 } ngx_http_charset_main_conf_t;
|
|
|
49
|
|
|
50
|
|
|
51 typedef struct {
|
|
206
|
52 ngx_int_t charset;
|
|
|
53 ngx_int_t source_charset;
|
|
|
54 ngx_flag_t override_charset;
|
|
50
|
55 } ngx_http_charset_loc_conf_t;
|
|
|
56
|
|
|
57
|
|
|
58 typedef struct {
|
|
206
|
59 u_char *table;
|
|
|
60 ngx_int_t charset;
|
|
|
61
|
|
|
62 ngx_chain_t *busy;
|
|
|
63 ngx_chain_t *free_bufs;
|
|
|
64 ngx_chain_t *free_buffers;
|
|
|
65
|
|
|
66 size_t saved_len;
|
|
|
67 u_char saved[NGX_UTF_LEN];
|
|
|
68
|
|
|
69 unsigned length:16;
|
|
|
70 unsigned from_utf8:1;
|
|
|
71 unsigned to_utf8:1;
|
|
50
|
72 } ngx_http_charset_ctx_t;
|
|
|
73
|
|
|
74
|
|
206
|
75 typedef struct {
|
|
|
76 ngx_http_charset_tables_t *table;
|
|
|
77 ngx_http_charset_t *charset;
|
|
|
78 ngx_uint_t characters;
|
|
|
79 } ngx_http_charset_conf_ctx_t;
|
|
|
80
|
|
50
|
81
|
|
206
|
82 static ngx_int_t ngx_http_charset_get_charset(ngx_http_charset_t *charsets,
|
|
244
|
83 ngx_uint_t n, ngx_str_t *charset);
|
|
206
|
84 static ngx_int_t ngx_http_charset_set_charset(ngx_http_request_t *r,
|
|
|
85 ngx_http_charset_t *charsets, ngx_int_t charset, ngx_int_t source_charset);
|
|
|
86 static ngx_uint_t ngx_http_charset_recode(ngx_buf_t *b, u_char *table);
|
|
|
87 static ngx_chain_t *ngx_http_charset_recode_from_utf8(ngx_pool_t *pool,
|
|
|
88 ngx_buf_t *buf, ngx_http_charset_ctx_t *ctx);
|
|
|
89 static ngx_chain_t *ngx_http_charset_recode_to_utf8(ngx_pool_t *pool,
|
|
|
90 ngx_buf_t *buf, ngx_http_charset_ctx_t *ctx);
|
|
|
91
|
|
|
92 static ngx_chain_t *ngx_http_charset_get_buf(ngx_pool_t *pool,
|
|
|
93 ngx_http_charset_ctx_t *ctx);
|
|
|
94 static ngx_chain_t *ngx_http_charset_get_buffer(ngx_pool_t *pool,
|
|
|
95 ngx_http_charset_ctx_t *ctx, size_t size);
|
|
|
96
|
|
|
97 static char *ngx_http_charset_map_block(ngx_conf_t *cf, ngx_command_t *cmd,
|
|
50
|
98 void *conf);
|
|
206
|
99 static char *ngx_http_charset_map(ngx_conf_t *cf, ngx_command_t *dummy,
|
|
|
100 void *conf);
|
|
50
|
101
|
|
|
102 static char *ngx_http_set_charset_slot(ngx_conf_t *cf, ngx_command_t *cmd,
|
|
|
103 void *conf);
|
|
|
104 static ngx_int_t ngx_http_add_charset(ngx_array_t *charsets, ngx_str_t *name);
|
|
|
105
|
|
|
106 static void *ngx_http_charset_create_main_conf(ngx_conf_t *cf);
|
|
|
107 static void *ngx_http_charset_create_loc_conf(ngx_conf_t *cf);
|
|
|
108 static char *ngx_http_charset_merge_loc_conf(ngx_conf_t *cf,
|
|
|
109 void *parent, void *child);
|
|
78
|
110 static ngx_int_t ngx_http_charset_postconfiguration(ngx_conf_t *cf);
|
|
50
|
111
|
|
|
112
|
|
|
113 static ngx_command_t ngx_http_charset_filter_commands[] = {
|
|
|
114
|
|
78
|
115 { ngx_string("charset"),
|
|
108
|
116 NGX_HTTP_MAIN_CONF|NGX_HTTP_SRV_CONF|NGX_HTTP_LOC_CONF
|
|
|
117 |NGX_HTTP_LIF_CONF|NGX_CONF_TAKE1,
|
|
78
|
118 ngx_http_set_charset_slot,
|
|
|
119 NGX_HTTP_LOC_CONF_OFFSET,
|
|
|
120 offsetof(ngx_http_charset_loc_conf_t, charset),
|
|
|
121 NULL },
|
|
|
122
|
|
|
123 { ngx_string("source_charset"),
|
|
108
|
124 NGX_HTTP_MAIN_CONF|NGX_HTTP_SRV_CONF|NGX_HTTP_LOC_CONF
|
|
|
125 |NGX_HTTP_LIF_CONF|NGX_CONF_TAKE1,
|
|
78
|
126 ngx_http_set_charset_slot,
|
|
|
127 NGX_HTTP_LOC_CONF_OFFSET,
|
|
|
128 offsetof(ngx_http_charset_loc_conf_t, source_charset),
|
|
|
129 NULL },
|
|
|
130
|
|
184
|
131 { ngx_string("override_charset"),
|
|
|
132 NGX_HTTP_MAIN_CONF|NGX_HTTP_SRV_CONF|NGX_HTTP_LOC_CONF
|
|
|
133 |NGX_HTTP_LIF_CONF|NGX_CONF_FLAG,
|
|
|
134 ngx_conf_set_flag_slot,
|
|
|
135 NGX_HTTP_LOC_CONF_OFFSET,
|
|
|
136 offsetof(ngx_http_charset_loc_conf_t, override_charset),
|
|
|
137 NULL },
|
|
|
138
|
|
50
|
139 { ngx_string("charset_map"),
|
|
|
140 NGX_HTTP_MAIN_CONF|NGX_CONF_BLOCK|NGX_CONF_TAKE2,
|
|
206
|
141 ngx_http_charset_map_block,
|
|
50
|
142 NGX_HTTP_MAIN_CONF_OFFSET,
|
|
|
143 0,
|
|
|
144 NULL },
|
|
|
145
|
|
|
146 ngx_null_command
|
|
|
147 };
|
|
|
148
|
|
|
149
|
|
|
150 static ngx_http_module_t ngx_http_charset_filter_module_ctx = {
|
|
58
|
151 NULL, /* preconfiguration */
|
|
78
|
152 ngx_http_charset_postconfiguration, /* postconfiguration */
|
|
50
|
153
|
|
|
154 ngx_http_charset_create_main_conf, /* create main configuration */
|
|
78
|
155 NULL, /* init main configuration */
|
|
50
|
156
|
|
|
157 NULL, /* create server configuration */
|
|
|
158 NULL, /* merge server configuration */
|
|
|
159
|
|
|
160 ngx_http_charset_create_loc_conf, /* create location configuration */
|
|
|
161 ngx_http_charset_merge_loc_conf /* merge location configuration */
|
|
|
162 };
|
|
|
163
|
|
|
164
|
|
|
165 ngx_module_t ngx_http_charset_filter_module = {
|
|
58
|
166 NGX_MODULE_V1,
|
|
50
|
167 &ngx_http_charset_filter_module_ctx, /* module context */
|
|
|
168 ngx_http_charset_filter_commands, /* module directives */
|
|
|
169 NGX_HTTP_MODULE, /* module type */
|
|
90
|
170 NULL, /* init master */
|
|
230
|
171 NULL, /* init module */
|
|
90
|
172 NULL, /* init process */
|
|
|
173 NULL, /* init thread */
|
|
|
174 NULL, /* exit thread */
|
|
|
175 NULL, /* exit process */
|
|
|
176 NULL, /* exit master */
|
|
|
177 NGX_MODULE_V1_PADDING
|
|
50
|
178 };
|
|
|
179
|
|
|
180
|
|
|
181 static ngx_http_output_header_filter_pt ngx_http_next_header_filter;
|
|
|
182 static ngx_http_output_body_filter_pt ngx_http_next_body_filter;
|
|
|
183
|
|
|
184
|
|
|
185 static ngx_int_t
|
|
|
186 ngx_http_charset_header_filter(ngx_http_request_t *r)
|
|
|
187 {
|
|
206
|
188 u_char *ct;
|
|
184
|
189 ngx_int_t charset, source_charset;
|
|
320
|
190 ngx_str_t *mc, *from, *to, s;
|
|
206
|
191 ngx_uint_t n;
|
|
50
|
192 ngx_http_charset_t *charsets;
|
|
|
193 ngx_http_charset_ctx_t *ctx;
|
|
244
|
194 ngx_http_variable_value_t *vv;
|
|
202
|
195 ngx_http_charset_loc_conf_t *lcf, *mlcf;
|
|
50
|
196 ngx_http_charset_main_conf_t *mcf;
|
|
|
197
|
|
|
198 mcf = ngx_http_get_module_main_conf(r, ngx_http_charset_filter_module);
|
|
184
|
199
|
|
206
|
200 charsets = mcf->charsets.elts;
|
|
|
201 n = mcf->charsets.nelts;
|
|
184
|
202
|
|
206
|
203 /* destination charset */
|
|
202
|
204
|
|
|
205 if (r == r->main) {
|
|
206
|
206
|
|
382
|
207 if (r->headers_out.content_encoding
|
|
|
208 && r->headers_out.content_encoding->value.len)
|
|
|
209 {
|
|
|
210 return ngx_http_next_header_filter(r);
|
|
|
211 }
|
|
|
212
|
|
202
|
213 if (r->headers_out.content_type.len == 0) {
|
|
|
214 return ngx_http_next_header_filter(r);
|
|
|
215 }
|
|
50
|
216
|
|
206
|
217 if (r->headers_out.override_charset
|
|
|
218 && r->headers_out.override_charset->len)
|
|
202
|
219 {
|
|
206
|
220 charset = ngx_http_charset_get_charset(charsets, n,
|
|
244
|
221 r->headers_out.override_charset);
|
|
206
|
222
|
|
|
223 if (charset == NGX_HTTP_NO_CHARSET) {
|
|
210
|
224 ngx_log_error(NGX_LOG_ERR, r->connection->log, 0,
|
|
206
|
225 "unknown charset \"%V\" to override",
|
|
208
|
226 r->headers_out.override_charset);
|
|
206
|
227
|
|
|
228 return ngx_http_next_header_filter(r);
|
|
|
229 }
|
|
|
230
|
|
|
231 } else {
|
|
|
232 mlcf = ngx_http_get_module_loc_conf(r,
|
|
|
233 ngx_http_charset_filter_module);
|
|
|
234 charset = mlcf->charset;
|
|
|
235
|
|
|
236 if (charset == NGX_HTTP_NO_CHARSET) {
|
|
|
237 return ngx_http_next_header_filter(r);
|
|
|
238 }
|
|
|
239
|
|
|
240 if (r->headers_out.charset.len) {
|
|
|
241 if (mlcf->override_charset == 0) {
|
|
|
242 return ngx_http_next_header_filter(r);
|
|
|
243 }
|
|
|
244
|
|
|
245 } else {
|
|
|
246 ct = r->headers_out.content_type.data;
|
|
|
247
|
|
286
|
248 if (ngx_strncasecmp(ct, (u_char *) "text/", 5) != 0
|
|
|
249 && ngx_strncasecmp(ct,
|
|
|
250 (u_char *) "application/x-javascript", 24)
|
|
|
251 != 0)
|
|
206
|
252 {
|
|
|
253 return ngx_http_next_header_filter(r);
|
|
|
254 }
|
|
|
255 }
|
|
244
|
256
|
|
|
257 if (charset >= NGX_HTTP_CHARSET_VAR) {
|
|
|
258 vv = ngx_http_get_indexed_variable(r,
|
|
|
259 charset - NGX_HTTP_CHARSET_VAR);
|
|
|
260
|
|
250
|
261 if (vv == NULL || vv->not_found) {
|
|
|
262 return NGX_ERROR;
|
|
|
263 }
|
|
|
264
|
|
320
|
265 s.len = vv->len;
|
|
|
266 s.data = vv->data;
|
|
|
267
|
|
|
268 charset = ngx_http_charset_get_charset(charsets, n, &s);
|
|
244
|
269 }
|
|
202
|
270 }
|
|
|
271
|
|
|
272 } else {
|
|
206
|
273 ctx = ngx_http_get_module_ctx(r->main, ngx_http_charset_filter_module);
|
|
|
274
|
|
|
275 if (ctx == NULL) {
|
|
|
276
|
|
|
277 mc = &r->main->headers_out.charset;
|
|
|
278
|
|
|
279 if (mc->len == 0) {
|
|
|
280 return ngx_http_next_header_filter(r);
|
|
|
281 }
|
|
|
282
|
|
|
283 ctx = ngx_pcalloc(r->pool, sizeof(ngx_http_charset_ctx_t));
|
|
|
284 if (ctx == NULL) {
|
|
|
285 return NGX_ERROR;
|
|
|
286 }
|
|
|
287
|
|
|
288 ngx_http_set_ctx(r->main, ctx, ngx_http_charset_filter_module);
|
|
202
|
289
|
|
244
|
290 charset = ngx_http_charset_get_charset(charsets, n, mc);
|
|
206
|
291
|
|
|
292 ctx->charset = charset;
|
|
|
293
|
|
210
|
294 } else {
|
|
|
295 charset = ctx->charset;
|
|
202
|
296 }
|
|
50
|
297 }
|
|
|
298
|
|
206
|
299 /* source charset */
|
|
184
|
300
|
|
206
|
301 if (r->headers_out.charset.len == 0) {
|
|
|
302 lcf = ngx_http_get_module_loc_conf(r, ngx_http_charset_filter_module);
|
|
184
|
303
|
|
244
|
304 source_charset = lcf->source_charset;
|
|
|
305
|
|
|
306 if (source_charset >= NGX_HTTP_CHARSET_VAR) {
|
|
|
307 vv = ngx_http_get_indexed_variable(r,
|
|
|
308 source_charset - NGX_HTTP_CHARSET_VAR);
|
|
|
309
|
|
250
|
310 if (vv == NULL || vv->not_found) {
|
|
|
311 return NGX_ERROR;
|
|
|
312 }
|
|
|
313
|
|
320
|
314 s.len = vv->len;
|
|
|
315 s.data = vv->data;
|
|
|
316
|
|
|
317 source_charset = ngx_http_charset_get_charset(charsets, n, &s);
|
|
244
|
318 }
|
|
|
319
|
|
210
|
320 if (charset != NGX_HTTP_NO_CHARSET) {
|
|
|
321 return ngx_http_charset_set_charset(r, mcf->charsets.elts, charset,
|
|
244
|
322 source_charset);
|
|
210
|
323 }
|
|
|
324
|
|
244
|
325 if (source_charset == NGX_CONF_UNSET) {
|
|
210
|
326 return ngx_http_next_header_filter(r);
|
|
|
327 }
|
|
|
328
|
|
244
|
329 from = &charsets[source_charset].name;
|
|
212
|
330 to = &r->main->headers_out.charset;
|
|
210
|
331
|
|
212
|
332 goto no_charset_map;
|
|
206
|
333 }
|
|
184
|
334
|
|
206
|
335 source_charset = ngx_http_charset_get_charset(charsets, n,
|
|
244
|
336 &r->headers_out.charset);
|
|
184
|
337
|
|
210
|
338 if (charset == NGX_HTTP_NO_CHARSET
|
|
|
339 || source_charset == NGX_HTTP_NO_CHARSET)
|
|
|
340 {
|
|
|
341 if (charset != source_charset
|
|
|
342 || ngx_strcasecmp(r->main->headers_out.charset.data,
|
|
|
343 r->headers_out.charset.data)
|
|
|
344 != 0)
|
|
|
345 {
|
|
212
|
346 from = &r->headers_out.charset;
|
|
|
347 to = (charset == NGX_HTTP_NO_CHARSET) ?
|
|
|
348 &r->main->headers_out.charset:
|
|
|
349 &charsets[charset].name;
|
|
|
350
|
|
|
351 goto no_charset_map;
|
|
210
|
352 }
|
|
184
|
353
|
|
206
|
354 return ngx_http_next_header_filter(r);
|
|
|
355 }
|
|
184
|
356
|
|
206
|
357 if (source_charset != charset
|
|
|
358 && (charsets[source_charset].tables == NULL
|
|
|
359 || charsets[source_charset].tables[charset] == NULL))
|
|
|
360 {
|
|
212
|
361 from = &charsets[source_charset].name;
|
|
|
362 to = &charsets[charset].name;
|
|
184
|
363
|
|
212
|
364 goto no_charset_map;
|
|
50
|
365 }
|
|
|
366
|
|
206
|
367 r->headers_out.content_type.len = r->headers_out.content_type_len;
|
|
|
368
|
|
|
369 return ngx_http_charset_set_charset(r, mcf->charsets.elts, charset,
|
|
|
370 source_charset);
|
|
212
|
371
|
|
|
372 no_charset_map:
|
|
|
373
|
|
|
374 ngx_log_error(NGX_LOG_ERR, r->connection->log, 0,
|
|
342
|
375 "no \"charset_map\" between the charsets \"%V\" and \"%V\"",
|
|
|
376 from, to);
|
|
212
|
377
|
|
|
378 return ngx_http_next_header_filter(r);
|
|
206
|
379 }
|
|
|
380
|
|
|
381
|
|
|
382 static ngx_int_t
|
|
|
383 ngx_http_charset_get_charset(ngx_http_charset_t *charsets, ngx_uint_t n,
|
|
244
|
384 ngx_str_t *charset)
|
|
206
|
385 {
|
|
|
386 ngx_uint_t i;
|
|
|
387
|
|
|
388 for (i = 0; i < n; i++) {
|
|
320
|
389 if (charsets[i].name.len != charset->len) {
|
|
244
|
390 continue;
|
|
|
391 }
|
|
|
392
|
|
320
|
393 if (ngx_strncasecmp(charsets[i].name.data, charset->data, charset->len)
|
|
|
394 == 0)
|
|
|
395 {
|
|
206
|
396 return i;
|
|
|
397 }
|
|
|
398 }
|
|
|
399
|
|
|
400 return NGX_HTTP_NO_CHARSET;
|
|
|
401 }
|
|
|
402
|
|
|
403
|
|
|
404 static ngx_int_t
|
|
|
405 ngx_http_charset_set_charset(ngx_http_request_t *r,
|
|
|
406 ngx_http_charset_t *charsets, ngx_int_t charset, ngx_int_t source_charset)
|
|
|
407 {
|
|
|
408 ngx_http_charset_ctx_t *ctx;
|
|
|
409
|
|
50
|
410 if (r->headers_out.status == NGX_HTTP_MOVED_PERMANENTLY
|
|
|
411 || r->headers_out.status == NGX_HTTP_MOVED_TEMPORARILY)
|
|
|
412 {
|
|
|
413 /*
|
|
184
|
414 * do not set charset for the redirect because NN 4.x
|
|
|
415 * use this charset instead of the next page charset
|
|
50
|
416 */
|
|
|
417
|
|
|
418 r->headers_out.charset.len = 0;
|
|
206
|
419
|
|
50
|
420 return ngx_http_next_header_filter(r);
|
|
|
421 }
|
|
|
422
|
|
184
|
423 r->headers_out.charset = charsets[charset].name;
|
|
|
424 r->utf8 = charsets[charset].utf8;
|
|
|
425
|
|
|
426 if (source_charset == NGX_CONF_UNSET || source_charset == charset) {
|
|
50
|
427 return ngx_http_next_header_filter(r);
|
|
|
428 }
|
|
|
429
|
|
|
430 ctx = ngx_pcalloc(r->pool, sizeof(ngx_http_charset_ctx_t));
|
|
|
431 if (ctx == NULL) {
|
|
|
432 return NGX_ERROR;
|
|
|
433 }
|
|
|
434
|
|
|
435 ngx_http_set_ctx(r, ctx, ngx_http_charset_filter_module);
|
|
|
436
|
|
184
|
437 ctx->table = charsets[source_charset].tables[charset];
|
|
|
438 ctx->charset = charset;
|
|
206
|
439 ctx->length = charsets[charset].length;
|
|
|
440 ctx->from_utf8 = charsets[source_charset].utf8;
|
|
|
441 ctx->to_utf8 = charsets[charset].utf8;
|
|
|
442
|
|
212
|
443 r->filter_need_in_memory = 1;
|
|
|
444
|
|
206
|
445 if ((ctx->to_utf8 || ctx->from_utf8) && r == r->main) {
|
|
|
446 ngx_http_clear_content_length(r);
|
|
212
|
447
|
|
|
448 } else {
|
|
|
449 r->filter_need_temporary = 1;
|
|
206
|
450 }
|
|
50
|
451
|
|
|
452 return ngx_http_next_header_filter(r);
|
|
|
453 }
|
|
|
454
|
|
|
455
|
|
|
456 static ngx_int_t
|
|
|
457 ngx_http_charset_body_filter(ngx_http_request_t *r, ngx_chain_t *in)
|
|
|
458 {
|
|
206
|
459 ngx_int_t rc;
|
|
|
460 ngx_buf_t *b;
|
|
|
461 ngx_chain_t *cl, *out, **ll;
|
|
184
|
462 ngx_http_charset_ctx_t *ctx;
|
|
50
|
463
|
|
|
464 ctx = ngx_http_get_module_ctx(r, ngx_http_charset_filter_module);
|
|
|
465
|
|
184
|
466 if (ctx == NULL || ctx->table == NULL) {
|
|
50
|
467 return ngx_http_next_body_filter(r, in);
|
|
|
468 }
|
|
|
469
|
|
206
|
470 if ((ctx->to_utf8 || ctx->from_utf8) || ctx->busy) {
|
|
|
471
|
|
|
472 out = NULL;
|
|
|
473 ll = &out;
|
|
|
474
|
|
|
475 for (cl = in; cl; cl = cl->next) {
|
|
|
476 b = cl->buf;
|
|
|
477
|
|
|
478 if (ngx_buf_size(b) == 0) {
|
|
214
|
479
|
|
|
480 *ll = ngx_alloc_chain_link(r->pool);
|
|
|
481 if (*ll == NULL) {
|
|
|
482 return NGX_ERROR;
|
|
|
483 }
|
|
|
484
|
|
|
485 (*ll)->buf = b;
|
|
|
486 (*ll)->next = NULL;
|
|
|
487
|
|
|
488 ll = &(*ll)->next;
|
|
|
489
|
|
206
|
490 continue;
|
|
|
491 }
|
|
|
492
|
|
|
493 if (ctx->to_utf8) {
|
|
|
494 *ll = ngx_http_charset_recode_to_utf8(r->pool, b, ctx);
|
|
|
495
|
|
|
496 } else {
|
|
|
497 *ll = ngx_http_charset_recode_from_utf8(r->pool, b, ctx);
|
|
|
498 }
|
|
|
499
|
|
|
500 if (*ll == NULL) {
|
|
|
501 return NGX_ERROR;
|
|
|
502 }
|
|
|
503
|
|
|
504 while (*ll) {
|
|
|
505 ll = &(*ll)->next;
|
|
|
506 }
|
|
|
507 }
|
|
|
508
|
|
|
509 rc = ngx_http_next_body_filter(r, out);
|
|
|
510
|
|
|
511 if (out) {
|
|
|
512 if (ctx->busy == NULL) {
|
|
|
513 ctx->busy = out;
|
|
|
514
|
|
|
515 } else {
|
|
|
516 for (cl = ctx->busy; cl->next; cl = cl->next) { /* void */ }
|
|
|
517 cl->next = out;
|
|
|
518 }
|
|
|
519 }
|
|
|
520
|
|
|
521 while (ctx->busy) {
|
|
|
522
|
|
|
523 cl = ctx->busy;
|
|
|
524 b = cl->buf;
|
|
|
525
|
|
|
526 if (ngx_buf_size(b) != 0) {
|
|
|
527 break;
|
|
|
528 }
|
|
|
529
|
|
|
530 #if (NGX_HAVE_WRITE_ZEROCOPY)
|
|
|
531 if (b->zerocopy_busy) {
|
|
|
532 break;
|
|
|
533 }
|
|
|
534 #endif
|
|
|
535
|
|
|
536 ctx->busy = cl->next;
|
|
|
537
|
|
|
538 if (b->tag != (ngx_buf_tag_t) &ngx_http_charset_filter_module) {
|
|
|
539 continue;
|
|
|
540 }
|
|
|
541
|
|
|
542 if (b->shadow) {
|
|
|
543 b->shadow->pos = b->shadow->last;
|
|
|
544 }
|
|
|
545
|
|
|
546 if (b->pos) {
|
|
|
547 cl->next = ctx->free_buffers;
|
|
|
548 ctx->free_buffers = cl;
|
|
|
549 continue;
|
|
|
550 }
|
|
|
551
|
|
|
552 cl->next = ctx->free_bufs;
|
|
|
553 ctx->free_bufs = cl;
|
|
|
554 }
|
|
|
555
|
|
|
556 return rc;
|
|
|
557 }
|
|
|
558
|
|
50
|
559 for (cl = in; cl; cl = cl->next) {
|
|
184
|
560 (void) ngx_http_charset_recode(cl->buf, ctx->table);
|
|
50
|
561 }
|
|
|
562
|
|
|
563 return ngx_http_next_body_filter(r, in);
|
|
|
564 }
|
|
|
565
|
|
|
566
|
|
|
567 static ngx_uint_t
|
|
184
|
568 ngx_http_charset_recode(ngx_buf_t *b, u_char *table)
|
|
50
|
569 {
|
|
370
|
570 u_char *p, *last;
|
|
184
|
571
|
|
370
|
572 last = b->last;
|
|
50
|
573
|
|
370
|
574 for (p = b->pos; p < last; p++) {
|
|
|
575
|
|
|
576 if (*p != table[*p]) {
|
|
|
577 goto recode;
|
|
50
|
578 }
|
|
|
579 }
|
|
|
580
|
|
184
|
581 return 0;
|
|
370
|
582
|
|
|
583 recode:
|
|
|
584
|
|
|
585 do {
|
|
|
586 if (*p != table[*p]) {
|
|
|
587 *p = table[*p];
|
|
|
588 }
|
|
|
589
|
|
|
590 p++;
|
|
|
591
|
|
|
592 } while (p < last);
|
|
|
593
|
|
|
594 b->in_file = 0;
|
|
|
595
|
|
|
596 return 1;
|
|
50
|
597 }
|
|
|
598
|
|
|
599
|
|
206
|
600 static ngx_chain_t *
|
|
|
601 ngx_http_charset_recode_from_utf8(ngx_pool_t *pool, ngx_buf_t *buf,
|
|
|
602 ngx_http_charset_ctx_t *ctx)
|
|
|
603 {
|
|
|
604 size_t len, size;
|
|
|
605 u_char c, *p, *src, *dst, *saved, **table;
|
|
|
606 uint32_t n;
|
|
|
607 ngx_buf_t *b;
|
|
|
608 ngx_uint_t i;
|
|
|
609 ngx_chain_t *out, *cl, **ll;
|
|
|
610
|
|
|
611 src = buf->pos;
|
|
|
612
|
|
|
613 if (ctx->saved_len == 0) {
|
|
|
614
|
|
|
615 for ( /* void */ ; src < buf->last; src++) {
|
|
|
616
|
|
|
617 if (*src < 0x80) {
|
|
|
618 continue;
|
|
|
619 }
|
|
|
620
|
|
|
621 len = src - buf->pos;
|
|
|
622
|
|
|
623 if (len > 512) {
|
|
|
624 out = ngx_http_charset_get_buf(pool, ctx);
|
|
|
625 if (out == NULL) {
|
|
|
626 return NULL;
|
|
|
627 }
|
|
|
628
|
|
|
629 b = out->buf;
|
|
|
630
|
|
|
631 b->temporary = buf->temporary;
|
|
|
632 b->memory = buf->memory;
|
|
|
633 b->mmap = buf->mmap;
|
|
|
634 b->flush = buf->flush;
|
|
|
635
|
|
|
636 b->pos = buf->pos;
|
|
|
637 b->last = src;
|
|
|
638
|
|
|
639 out->buf = b;
|
|
|
640 out->next = NULL;
|
|
|
641
|
|
|
642 size = buf->last - src;
|
|
|
643
|
|
|
644 saved = src;
|
|
|
645 n = ngx_utf_decode(&saved, size);
|
|
|
646
|
|
|
647 if (n == 0xfffffffe) {
|
|
|
648 /* incomplete UTF-8 symbol */
|
|
|
649
|
|
|
650 ngx_memcpy(ctx->saved, src, size);
|
|
|
651 ctx->saved_len = size;
|
|
|
652
|
|
|
653 b->shadow = buf;
|
|
|
654
|
|
|
655 return out;
|
|
|
656 }
|
|
|
657
|
|
|
658 } else {
|
|
|
659 out = NULL;
|
|
|
660 size = len + buf->last - src;
|
|
|
661 src = buf->pos;
|
|
|
662 }
|
|
|
663
|
|
|
664 if (size < NGX_HTML_ENTITY_LEN) {
|
|
|
665 size += NGX_HTML_ENTITY_LEN;
|
|
|
666 }
|
|
|
667
|
|
|
668 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
|
669 if (cl == NULL) {
|
|
|
670 return NULL;
|
|
|
671 }
|
|
|
672
|
|
|
673 if (out) {
|
|
|
674 out->next = cl;
|
|
|
675
|
|
|
676 } else {
|
|
|
677 out = cl;
|
|
|
678 }
|
|
|
679
|
|
|
680 b = cl->buf;
|
|
|
681 dst = b->pos;
|
|
|
682
|
|
|
683 goto recode;
|
|
|
684 }
|
|
|
685
|
|
|
686 out = ngx_alloc_chain_link(pool);
|
|
|
687 if (out == NULL) {
|
|
|
688 return NULL;
|
|
|
689 }
|
|
|
690
|
|
|
691 out->buf = buf;
|
|
|
692 out->next = NULL;
|
|
|
693
|
|
|
694 return out;
|
|
|
695 }
|
|
|
696
|
|
|
697 /* process incomplete UTF sequence from previous buffer */
|
|
|
698
|
|
|
699 ngx_log_debug1(NGX_LOG_DEBUG_HTTP, pool->log, 0,
|
|
|
700 "http charset utf saved: %z", ctx->saved_len);
|
|
|
701
|
|
|
702 p = src;
|
|
|
703
|
|
|
704 for (i = ctx->saved_len; i < NGX_UTF_LEN; i++) {
|
|
|
705 ctx->saved[i] = *p++;
|
|
|
706
|
|
|
707 if (p == buf->last) {
|
|
|
708 break;
|
|
|
709 }
|
|
|
710 }
|
|
|
711
|
|
|
712 saved = ctx->saved;
|
|
|
713 n = ngx_utf_decode(&saved, i);
|
|
|
714
|
|
|
715 c = '\0';
|
|
|
716
|
|
|
717 if (n < 0x10000) {
|
|
|
718 table = (u_char **) ctx->table;
|
|
|
719 p = table[n >> 8];
|
|
|
720
|
|
|
721 if (p) {
|
|
|
722 c = p[n & 0xff];
|
|
|
723 }
|
|
|
724
|
|
|
725 } else if (n == 0xfffffffe) {
|
|
|
726
|
|
|
727 /* incomplete UTF-8 symbol */
|
|
|
728
|
|
|
729 if (i < NGX_UTF_LEN) {
|
|
|
730 out = ngx_http_charset_get_buf(pool, ctx);
|
|
|
731 if (out == NULL) {
|
|
|
732 return NULL;
|
|
|
733 }
|
|
|
734
|
|
|
735 b = out->buf;
|
|
|
736
|
|
|
737 b->pos = buf->pos;
|
|
|
738 b->last = buf->last;
|
|
|
739 b->sync = 1;
|
|
|
740 b->shadow = buf;
|
|
|
741
|
|
|
742 ngx_memcpy(&ctx->saved[ctx->saved_len], src, i);
|
|
|
743 ctx->saved_len += i;
|
|
|
744
|
|
|
745 return out;
|
|
|
746 }
|
|
|
747 }
|
|
|
748
|
|
|
749 size = buf->last - buf->pos;
|
|
|
750
|
|
|
751 if (size < NGX_HTML_ENTITY_LEN) {
|
|
|
752 size += NGX_HTML_ENTITY_LEN;
|
|
|
753 }
|
|
|
754
|
|
|
755 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
|
756 if (cl == NULL) {
|
|
|
757 return NULL;
|
|
|
758 }
|
|
|
759
|
|
|
760 out = cl;
|
|
|
761
|
|
|
762 b = cl->buf;
|
|
|
763 dst = b->pos;
|
|
|
764
|
|
|
765 if (c) {
|
|
|
766 *dst++ = c;
|
|
|
767
|
|
|
768 } else if (n == 0xfffffffe) {
|
|
|
769 *dst++ = '?';
|
|
|
770
|
|
|
771 ngx_log_debug0(NGX_LOG_DEBUG_HTTP, pool->log, 0,
|
|
|
772 "http charset invalid utf 0");
|
|
|
773
|
|
|
774 saved = &ctx->saved[NGX_UTF_LEN];
|
|
|
775
|
|
|
776 } else if (n > 0x10ffff) {
|
|
|
777 *dst++ = '?';
|
|
|
778
|
|
|
779 ngx_log_debug0(NGX_LOG_DEBUG_HTTP, pool->log, 0,
|
|
|
780 "http charset invalid utf 1");
|
|
|
781
|
|
|
782 } else {
|
|
|
783 dst = ngx_sprintf(dst, "&#%uD;", n);
|
|
|
784 }
|
|
|
785
|
|
|
786 src += (saved - ctx->saved) - ctx->saved_len;
|
|
|
787 ctx->saved_len = 0;
|
|
|
788
|
|
|
789 recode:
|
|
|
790
|
|
|
791 ll = &cl->next;
|
|
|
792
|
|
|
793 table = (u_char **) ctx->table;
|
|
|
794
|
|
|
795 while (src < buf->last) {
|
|
|
796
|
|
|
797 if ((size_t) (b->end - dst) < NGX_HTML_ENTITY_LEN) {
|
|
|
798 b->last = dst;
|
|
|
799
|
|
|
800 size = buf->last - src + NGX_HTML_ENTITY_LEN;
|
|
|
801
|
|
|
802 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
|
803 if (cl == NULL) {
|
|
|
804 return NULL;
|
|
|
805 }
|
|
|
806
|
|
|
807 *ll = cl;
|
|
|
808 ll = &cl->next;
|
|
|
809
|
|
|
810 b = cl->buf;
|
|
|
811 dst = b->pos;
|
|
|
812 }
|
|
|
813
|
|
|
814 if (*src < 0x80) {
|
|
|
815 *dst++ = *src++;
|
|
|
816 continue;
|
|
|
817 }
|
|
|
818
|
|
|
819 len = buf->last - src;
|
|
|
820
|
|
|
821 n = ngx_utf_decode(&src, len);
|
|
|
822
|
|
|
823 if (n < 0x10000) {
|
|
|
824
|
|
|
825 p = table[n >> 8];
|
|
|
826
|
|
|
827 if (p) {
|
|
|
828 c = p[n & 0xff];
|
|
|
829
|
|
|
830 if (c) {
|
|
|
831 *dst++ = c;
|
|
|
832 continue;
|
|
|
833 }
|
|
|
834 }
|
|
|
835
|
|
|
836 dst = ngx_sprintf(dst, "&#%uD;", n);
|
|
|
837
|
|
|
838 continue;
|
|
|
839 }
|
|
|
840
|
|
|
841 if (n == 0xfffffffe) {
|
|
|
842 /* incomplete UTF-8 symbol */
|
|
|
843
|
|
|
844 ngx_memcpy(ctx->saved, src, len);
|
|
|
845 ctx->saved_len = len;
|
|
|
846
|
|
|
847 if (b->pos == dst) {
|
|
|
848 b->sync = 1;
|
|
|
849 b->temporary = 0;
|
|
|
850 }
|
|
|
851
|
|
|
852 break;
|
|
|
853 }
|
|
|
854
|
|
|
855 if (n > 0x10ffff) {
|
|
|
856 *dst++ = '?';
|
|
|
857
|
|
|
858 ngx_log_debug0(NGX_LOG_DEBUG_HTTP, pool->log, 0,
|
|
|
859 "http charset invalid utf 2");
|
|
|
860
|
|
|
861 continue;
|
|
|
862 }
|
|
|
863
|
|
|
864 /* n > 0xffff */
|
|
|
865
|
|
|
866 dst = ngx_sprintf(dst, "&#%uD;", n);
|
|
|
867 }
|
|
|
868
|
|
|
869 b->last = dst;
|
|
|
870
|
|
|
871 b->last_buf = buf->last_buf;
|
|
|
872 b->last_in_chain = buf->last_in_chain;
|
|
|
873 b->flush = buf->flush;
|
|
|
874
|
|
|
875 b->shadow = buf;
|
|
|
876
|
|
|
877 return out;
|
|
|
878 }
|
|
|
879
|
|
|
880
|
|
|
881 static ngx_chain_t *
|
|
|
882 ngx_http_charset_recode_to_utf8(ngx_pool_t *pool, ngx_buf_t *buf,
|
|
|
883 ngx_http_charset_ctx_t *ctx)
|
|
|
884 {
|
|
|
885 size_t len, size;
|
|
|
886 u_char *p, *src, *dst, *table;
|
|
|
887 ngx_buf_t *b;
|
|
|
888 ngx_chain_t *out, *cl, **ll;
|
|
|
889
|
|
|
890 table = ctx->table;
|
|
|
891
|
|
|
892 for (src = buf->pos; src < buf->last; src++) {
|
|
|
893 if (table[*src * NGX_UTF_LEN] == '\1') {
|
|
|
894 continue;
|
|
|
895 }
|
|
|
896
|
|
|
897 goto recode;
|
|
|
898 }
|
|
|
899
|
|
|
900 out = ngx_alloc_chain_link(pool);
|
|
|
901 if (out == NULL) {
|
|
|
902 return NULL;
|
|
|
903 }
|
|
|
904
|
|
|
905 out->buf = buf;
|
|
|
906 out->next = NULL;
|
|
|
907
|
|
|
908 return out;
|
|
|
909
|
|
|
910 recode:
|
|
|
911
|
|
|
912 /*
|
|
|
913 * we assume that there are about half of characters to be recoded,
|
|
|
914 * so we preallocate "size / 2 + size / 2 * ctx->length"
|
|
|
915 */
|
|
|
916
|
|
|
917 len = src - buf->pos;
|
|
|
918
|
|
|
919 if (len > 512) {
|
|
|
920 out = ngx_http_charset_get_buf(pool, ctx);
|
|
|
921 if (out == NULL) {
|
|
|
922 return NULL;
|
|
|
923 }
|
|
|
924
|
|
|
925 b = out->buf;
|
|
|
926
|
|
|
927 b->temporary = buf->temporary;
|
|
|
928 b->memory = buf->memory;
|
|
|
929 b->mmap = buf->mmap;
|
|
|
930 b->flush = buf->flush;
|
|
|
931
|
|
|
932 b->pos = buf->pos;
|
|
|
933 b->last = src;
|
|
|
934
|
|
|
935 out->buf = b;
|
|
|
936 out->next = NULL;
|
|
|
937
|
|
|
938 size = buf->last - src;
|
|
|
939 size = size / 2 + size / 2 * ctx->length;
|
|
|
940
|
|
|
941 } else {
|
|
|
942 out = NULL;
|
|
|
943
|
|
|
944 size = buf->last - src;
|
|
|
945 size = len + size / 2 + size / 2 * ctx->length;
|
|
|
946
|
|
|
947 src = buf->pos;
|
|
|
948 }
|
|
|
949
|
|
|
950 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
|
951 if (cl == NULL) {
|
|
|
952 return NULL;
|
|
|
953 }
|
|
|
954
|
|
|
955 if (out) {
|
|
|
956 out->next = cl;
|
|
|
957
|
|
|
958 } else {
|
|
|
959 out = cl;
|
|
|
960 }
|
|
|
961
|
|
|
962 ll = &cl->next;
|
|
|
963
|
|
|
964 b = cl->buf;
|
|
|
965 dst = b->pos;
|
|
|
966
|
|
|
967 while (src < buf->last) {
|
|
|
968
|
|
|
969 p = &table[*src++ * NGX_UTF_LEN];
|
|
|
970 len = *p++;
|
|
|
971
|
|
|
972 if ((size_t) (b->end - dst) < len) {
|
|
|
973 b->last = dst;
|
|
|
974
|
|
|
975 size = buf->last - src;
|
|
|
976 size = len + size / 2 + size / 2 * ctx->length;
|
|
|
977
|
|
|
978 cl = ngx_http_charset_get_buffer(pool, ctx, size);
|
|
|
979 if (cl == NULL) {
|
|
|
980 return NULL;
|
|
|
981 }
|
|
|
982
|
|
|
983 *ll = cl;
|
|
|
984 ll = &cl->next;
|
|
|
985
|
|
|
986 b = cl->buf;
|
|
|
987 dst = b->pos;
|
|
|
988 }
|
|
|
989
|
|
|
990 while (len) {
|
|
|
991 *dst++ = *p++;
|
|
|
992 len--;
|
|
|
993 }
|
|
|
994 }
|
|
|
995
|
|
|
996 b->last = dst;
|
|
|
997
|
|
|
998 b->last_buf = buf->last_buf;
|
|
|
999 b->last_in_chain = buf->last_in_chain;
|
|
|
1000 b->flush = buf->flush;
|
|
|
1001
|
|
|
1002 b->shadow = buf;
|
|
|
1003
|
|
|
1004 return out;
|
|
|
1005 }
|
|
|
1006
|
|
|
1007
|
|
|
1008 static ngx_chain_t *
|
|
|
1009 ngx_http_charset_get_buf(ngx_pool_t *pool, ngx_http_charset_ctx_t *ctx)
|
|
|
1010 {
|
|
|
1011 ngx_chain_t *cl;
|
|
|
1012
|
|
|
1013 cl = ctx->free_bufs;
|
|
|
1014
|
|
|
1015 if (cl) {
|
|
|
1016 ctx->free_bufs = cl->next;
|
|
|
1017
|
|
|
1018 cl->buf->shadow = NULL;
|
|
|
1019 cl->next = NULL;
|
|
|
1020
|
|
|
1021 return cl;
|
|
|
1022 }
|
|
|
1023
|
|
|
1024 cl = ngx_alloc_chain_link(pool);
|
|
|
1025 if (cl == NULL) {
|
|
|
1026 return NULL;
|
|
|
1027 }
|
|
|
1028
|
|
|
1029 cl->buf = ngx_calloc_buf(pool);
|
|
|
1030 if (cl->buf == NULL) {
|
|
|
1031 return NULL;
|
|
|
1032 }
|
|
|
1033
|
|
|
1034 cl->next = NULL;
|
|
|
1035
|
|
|
1036 cl->buf->tag = (ngx_buf_tag_t) &ngx_http_charset_filter_module;
|
|
|
1037
|
|
|
1038 return cl;
|
|
|
1039 }
|
|
|
1040
|
|
|
1041
|
|
|
1042 static ngx_chain_t *
|
|
|
1043 ngx_http_charset_get_buffer(ngx_pool_t *pool, ngx_http_charset_ctx_t *ctx,
|
|
|
1044 size_t size)
|
|
|
1045 {
|
|
|
1046 ngx_buf_t *b;
|
|
|
1047 ngx_chain_t *cl, **ll;
|
|
|
1048
|
|
|
1049 for (ll = &ctx->free_buffers, cl = ctx->free_buffers;
|
|
|
1050 cl;
|
|
|
1051 ll = &cl->next, cl = cl->next)
|
|
|
1052 {
|
|
|
1053 b = cl->buf;
|
|
|
1054
|
|
|
1055 if ((size_t) (b->end - b->start) >= size) {
|
|
|
1056 *ll = cl->next;
|
|
|
1057 cl->next = NULL;
|
|
|
1058
|
|
|
1059 b->pos = b->start;
|
|
|
1060 b->temporary = 1;
|
|
|
1061 b->shadow = NULL;
|
|
|
1062
|
|
|
1063 return cl;
|
|
|
1064 }
|
|
|
1065 }
|
|
|
1066
|
|
|
1067 cl = ngx_alloc_chain_link(pool);
|
|
|
1068 if (cl == NULL) {
|
|
|
1069 return NULL;
|
|
|
1070 }
|
|
|
1071
|
|
|
1072 cl->buf = ngx_create_temp_buf(pool, size);
|
|
|
1073 if (cl->buf == NULL) {
|
|
|
1074 return NULL;
|
|
|
1075 }
|
|
|
1076
|
|
|
1077 cl->next = NULL;
|
|
|
1078
|
|
|
1079 cl->buf->temporary = 1;
|
|
|
1080 cl->buf->tag = (ngx_buf_tag_t) &ngx_http_charset_filter_module;
|
|
|
1081
|
|
|
1082 return cl;
|
|
|
1083 }
|
|
|
1084
|
|
|
1085
|
|
50
|
1086 static char *
|
|
206
|
1087 ngx_http_charset_map_block(ngx_conf_t *cf, ngx_command_t *cmd, void *conf)
|
|
50
|
1088 {
|
|
|
1089 ngx_http_charset_main_conf_t *mcf = conf;
|
|
|
1090
|
|
206
|
1091 char *rv;
|
|
|
1092 u_char *p, *dst2src, **pp;
|
|
|
1093 ngx_int_t src, dst;
|
|
|
1094 ngx_uint_t i, n;
|
|
|
1095 ngx_str_t *value;
|
|
|
1096 ngx_conf_t pvcf;
|
|
|
1097 ngx_http_charset_t *charset;
|
|
|
1098 ngx_http_charset_tables_t *table;
|
|
|
1099 ngx_http_charset_conf_ctx_t ctx;
|
|
50
|
1100
|
|
|
1101 value = cf->args->elts;
|
|
|
1102
|
|
|
1103 src = ngx_http_add_charset(&mcf->charsets, &value[1]);
|
|
|
1104 if (src == NGX_ERROR) {
|
|
|
1105 return NGX_CONF_ERROR;
|
|
|
1106 }
|
|
|
1107
|
|
|
1108 dst = ngx_http_add_charset(&mcf->charsets, &value[2]);
|
|
|
1109 if (dst == NGX_ERROR) {
|
|
|
1110 return NGX_CONF_ERROR;
|
|
|
1111 }
|
|
|
1112
|
|
|
1113 if (src == dst) {
|
|
|
1114 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
|
1115 "\"charset_map\" between the same charsets "
|
|
|
1116 "\"%V\" and \"%V\"", &value[1], &value[2]);
|
|
|
1117 return NGX_CONF_ERROR;
|
|
|
1118 }
|
|
|
1119
|
|
|
1120 table = mcf->tables.elts;
|
|
|
1121 for (i = 0; i < mcf->tables.nelts; i++) {
|
|
|
1122 if ((src == table->src && dst == table->dst)
|
|
|
1123 || (src == table->dst && dst == table->src))
|
|
|
1124 {
|
|
|
1125 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
|
1126 "duplicate \"charset_map\" between "
|
|
|
1127 "\"%V\" and \"%V\"", &value[1], &value[2]);
|
|
|
1128 return NGX_CONF_ERROR;
|
|
|
1129 }
|
|
|
1130 }
|
|
|
1131
|
|
|
1132 table = ngx_array_push(&mcf->tables);
|
|
|
1133 if (table == NULL) {
|
|
|
1134 return NGX_CONF_ERROR;
|
|
|
1135 }
|
|
|
1136
|
|
|
1137 table->src = src;
|
|
|
1138 table->dst = dst;
|
|
|
1139
|
|
286
|
1140 if (ngx_strcasecmp(value[2].data, (u_char *) "utf-8") == 0) {
|
|
206
|
1141 table->src2dst = ngx_pcalloc(cf->pool, 256 * NGX_UTF_LEN);
|
|
|
1142 if (table->src2dst == NULL) {
|
|
|
1143 return NGX_CONF_ERROR;
|
|
|
1144 }
|
|
|
1145
|
|
|
1146 table->dst2src = ngx_pcalloc(cf->pool, 256 * sizeof(void *));
|
|
|
1147 if (table->dst2src == NULL) {
|
|
|
1148 return NGX_CONF_ERROR;
|
|
|
1149 }
|
|
|
1150
|
|
|
1151 dst2src = ngx_pcalloc(cf->pool, 256);
|
|
|
1152 if (dst2src == NULL) {
|
|
|
1153 return NGX_CONF_ERROR;
|
|
|
1154 }
|
|
|
1155
|
|
|
1156 pp = (u_char **) &table->dst2src[0];
|
|
|
1157 pp[0] = dst2src;
|
|
|
1158
|
|
|
1159 for (i = 0; i < 128; i++) {
|
|
|
1160 p = &table->src2dst[i * NGX_UTF_LEN];
|
|
|
1161 p[0] = '\1';
|
|
|
1162 p[1] = (u_char) i;
|
|
|
1163 dst2src[i] = (u_char) i;
|
|
|
1164 }
|
|
50
|
1165
|
|
206
|
1166 for (/* void */; i < 256; i++) {
|
|
|
1167 p = &table->src2dst[i * NGX_UTF_LEN];
|
|
|
1168 p[0] = '\1';
|
|
|
1169 p[1] = '?';
|
|
|
1170 }
|
|
|
1171
|
|
|
1172 } else {
|
|
|
1173 table->src2dst = ngx_palloc(cf->pool, 256);
|
|
|
1174 if (table->src2dst == NULL) {
|
|
|
1175 return NGX_CONF_ERROR;
|
|
|
1176 }
|
|
|
1177
|
|
|
1178 table->dst2src = ngx_palloc(cf->pool, 256);
|
|
|
1179 if (table->dst2src == NULL) {
|
|
|
1180 return NGX_CONF_ERROR;
|
|
|
1181 }
|
|
|
1182
|
|
|
1183 for (i = 0; i < 128; i++) {
|
|
|
1184 table->src2dst[i] = (u_char) i;
|
|
|
1185 table->dst2src[i] = (u_char) i;
|
|
|
1186 }
|
|
|
1187
|
|
|
1188 for (/* void */; i < 256; i++) {
|
|
|
1189 table->src2dst[i] = '?';
|
|
|
1190 table->dst2src[i] = '?';
|
|
|
1191 }
|
|
50
|
1192 }
|
|
|
1193
|
|
206
|
1194 charset = mcf->charsets.elts;
|
|
50
|
1195
|
|
206
|
1196 ctx.table = table;
|
|
|
1197 ctx.charset = &charset[dst];
|
|
|
1198 ctx.characters = 0;
|
|
50
|
1199
|
|
|
1200 pvcf = *cf;
|
|
206
|
1201 cf->ctx = &ctx;
|
|
|
1202 cf->handler = ngx_http_charset_map;
|
|
50
|
1203 cf->handler_conf = conf;
|
|
|
1204
|
|
|
1205 rv = ngx_conf_parse(cf, NULL);
|
|
|
1206
|
|
|
1207 *cf = pvcf;
|
|
|
1208
|
|
206
|
1209 if (ctx.characters) {
|
|
|
1210 n = ctx.charset->length;
|
|
|
1211 ctx.charset->length /= ctx.characters;
|
|
|
1212
|
|
|
1213 if (((n * 10) / ctx.characters) % 10 > 4) {
|
|
|
1214 ctx.charset->length++;
|
|
|
1215 }
|
|
|
1216 }
|
|
|
1217
|
|
50
|
1218 return rv;
|
|
|
1219 }
|
|
|
1220
|
|
|
1221
|
|
|
1222 static char *
|
|
206
|
1223 ngx_http_charset_map(ngx_conf_t *cf, ngx_command_t *dummy, void *conf)
|
|
50
|
1224 {
|
|
206
|
1225 u_char *p, *dst2src, **pp;
|
|
|
1226 uint32_t n;
|
|
|
1227 ngx_int_t src, dst;
|
|
|
1228 ngx_str_t *value;
|
|
|
1229 ngx_uint_t i;
|
|
|
1230 ngx_http_charset_tables_t *table;
|
|
|
1231 ngx_http_charset_conf_ctx_t *ctx;
|
|
50
|
1232
|
|
|
1233 if (cf->args->nelts != 2) {
|
|
|
1234 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0, "invalid parameters number");
|
|
|
1235 return NGX_CONF_ERROR;
|
|
|
1236 }
|
|
|
1237
|
|
|
1238 value = cf->args->elts;
|
|
|
1239
|
|
|
1240 src = ngx_hextoi(value[0].data, value[0].len);
|
|
|
1241 if (src == NGX_ERROR || src > 255) {
|
|
|
1242 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
|
1243 "invalid value \"%V\"", &value[0]);
|
|
|
1244 return NGX_CONF_ERROR;
|
|
|
1245 }
|
|
|
1246
|
|
206
|
1247 ctx = cf->ctx;
|
|
|
1248 table = ctx->table;
|
|
|
1249
|
|
|
1250 if (ctx->charset->utf8) {
|
|
|
1251 p = &table->src2dst[src * NGX_UTF_LEN];
|
|
|
1252
|
|
|
1253 *p++ = (u_char) (value[1].len / 2);
|
|
|
1254
|
|
|
1255 for (i = 0; i < value[1].len; i += 2) {
|
|
|
1256 dst = ngx_hextoi(&value[1].data[i], 2);
|
|
|
1257 if (dst == NGX_ERROR || dst > 255) {
|
|
|
1258 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
|
1259 "invalid value \"%V\"", &value[1]);
|
|
|
1260 return NGX_CONF_ERROR;
|
|
|
1261 }
|
|
|
1262
|
|
|
1263 *p++ = (u_char) dst;
|
|
|
1264 }
|
|
|
1265
|
|
|
1266 i /= 2;
|
|
|
1267
|
|
|
1268 ctx->charset->length += i;
|
|
|
1269 ctx->characters++;
|
|
|
1270
|
|
|
1271 p = &table->src2dst[src * NGX_UTF_LEN] + 1;
|
|
|
1272
|
|
|
1273 n = ngx_utf_decode(&p, i);
|
|
|
1274
|
|
|
1275 if (n > 0xffff) {
|
|
|
1276 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
|
1277 "invalid value \"%V\"", &value[1]);
|
|
|
1278 return NGX_CONF_ERROR;
|
|
|
1279 }
|
|
|
1280
|
|
|
1281 pp = (u_char **) &table->dst2src[0];
|
|
|
1282
|
|
|
1283 dst2src = pp[n >> 8];
|
|
|
1284
|
|
|
1285 if (dst2src == NULL) {
|
|
|
1286 dst2src = ngx_pcalloc(cf->pool, 256);
|
|
|
1287 if (dst2src == NULL) {
|
|
|
1288 return NGX_CONF_ERROR;
|
|
|
1289 }
|
|
|
1290
|
|
|
1291 pp[n >> 8] = dst2src;
|
|
|
1292 }
|
|
|
1293
|
|
|
1294 dst2src[n & 0xff] = (u_char) src;
|
|
|
1295
|
|
|
1296 } else {
|
|
|
1297 dst = ngx_hextoi(value[1].data, value[1].len);
|
|
|
1298 if (dst == NGX_ERROR || dst > 255) {
|
|
|
1299 ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
|
|
|
1300 "invalid value \"%V\"", &value[1]);
|
|
|
1301 return NGX_CONF_ERROR;
|
|
|
1302 }
|
|
|
1303
|
|
|
1304 table->src2dst[src] = (u_char) dst;
|
|
|
1305 table->dst2src[dst] = (u_char) src;
|
|
50
|
1306 }
|
|
|
1307
|
|
|
1308 return NGX_CONF_OK;
|
|
|
1309 }
|
|
|
1310
|
|
|
1311
|
|
|
1312 static char *
|
|
|
1313 ngx_http_set_charset_slot(ngx_conf_t *cf, ngx_command_t *cmd, void *conf)
|
|
|
1314 {
|
|
|
1315 char *p = conf;
|
|
|
1316
|
|
|
1317 ngx_int_t *cp;
|
|
244
|
1318 ngx_str_t *value, var;
|
|
50
|
1319 ngx_http_charset_main_conf_t *mcf;
|
|
|
1320
|
|
|
1321 cp = (ngx_int_t *) (p + cmd->offset);
|
|
|
1322
|
|
|
1323 if (*cp != NGX_CONF_UNSET) {
|
|
|
1324 return "is duplicate";
|
|
|
1325 }
|
|
|
1326
|
|
78
|
1327 value = cf->args->elts;
|
|
|
1328
|
|
|
1329 if (cmd->offset == offsetof(ngx_http_charset_loc_conf_t, charset)
|
|
|
1330 && ngx_strcmp(value[1].data, "off") == 0)
|
|
|
1331 {
|
|
|
1332 *cp = NGX_HTTP_NO_CHARSET;
|
|
|
1333 return NGX_CONF_OK;
|
|
|
1334 }
|
|
|
1335
|
|
244
|
1336
|
|
|
1337 if (value[1].data[0] == '$') {
|
|
|
1338 var.len = value[1].len - 1;
|
|
|
1339 var.data = value[1].data + 1;
|
|
|
1340
|
|
|
1341 *cp = ngx_http_get_variable_index(cf, &var);
|
|
|
1342
|
|
|
1343 if (*cp == NGX_ERROR) {
|
|
|
1344 return NGX_CONF_ERROR;
|
|
|
1345 }
|
|
|
1346
|
|
|
1347 *cp += NGX_HTTP_CHARSET_VAR;
|
|
|
1348
|
|
|
1349 return NGX_CONF_OK;
|
|
|
1350 }
|
|
|
1351
|
|
50
|
1352 mcf = ngx_http_conf_get_module_main_conf(cf,
|
|
|
1353 ngx_http_charset_filter_module);
|
|
|
1354
|
|
|
1355 *cp = ngx_http_add_charset(&mcf->charsets, &value[1]);
|
|
|
1356 if (*cp == NGX_ERROR) {
|
|
|
1357 return NGX_CONF_ERROR;
|
|
|
1358 }
|
|
|
1359
|
|
|
1360 return NGX_CONF_OK;
|
|
|
1361 }
|
|
|
1362
|
|
|
1363
|
|
|
1364 static ngx_int_t
|
|
|
1365 ngx_http_add_charset(ngx_array_t *charsets, ngx_str_t *name)
|
|
|
1366 {
|
|
|
1367 ngx_uint_t i;
|
|
|
1368 ngx_http_charset_t *c;
|
|
|
1369
|
|
|
1370 c = charsets->elts;
|
|
|
1371 for (i = 0; i < charsets->nelts; i++) {
|
|
|
1372 if (name->len != c[i].name.len) {
|
|
|
1373 continue;
|
|
|
1374 }
|
|
|
1375
|
|
|
1376 if (ngx_strcasecmp(name->data, c[i].name.data) == 0) {
|
|
|
1377 break;
|
|
|
1378 }
|
|
|
1379 }
|
|
|
1380
|
|
|
1381 if (i < charsets->nelts) {
|
|
|
1382 return i;
|
|
|
1383 }
|
|
|
1384
|
|
|
1385 c = ngx_array_push(charsets);
|
|
|
1386 if (c == NULL) {
|
|
|
1387 return NGX_ERROR;
|
|
|
1388 }
|
|
|
1389
|
|
|
1390 c->tables = NULL;
|
|
|
1391 c->name = *name;
|
|
206
|
1392 c->length = 0;
|
|
50
|
1393
|
|
286
|
1394 if (ngx_strcasecmp(name->data, (u_char *) "utf-8") == 0) {
|
|
72
|
1395 c->utf8 = 1;
|
|
216
|
1396
|
|
|
1397 } else {
|
|
|
1398 c->utf8 = 0;
|
|
72
|
1399 }
|
|
|
1400
|
|
50
|
1401 return i;
|
|
|
1402 }
|
|
|
1403
|
|
|
1404
|
|
|
1405 static void *
|
|
|
1406 ngx_http_charset_create_main_conf(ngx_conf_t *cf)
|
|
|
1407 {
|
|
|
1408 ngx_http_charset_main_conf_t *mcf;
|
|
|
1409
|
|
|
1410 mcf = ngx_pcalloc(cf->pool, sizeof(ngx_http_charset_main_conf_t));
|
|
|
1411 if (mcf == NULL) {
|
|
|
1412 return NGX_CONF_ERROR;
|
|
|
1413 }
|
|
|
1414
|
|
|
1415 if (ngx_array_init(&mcf->charsets, cf->pool, 2, sizeof(ngx_http_charset_t))
|
|
78
|
1416 == NGX_ERROR)
|
|
50
|
1417 {
|
|
|
1418 return NGX_CONF_ERROR;
|
|
|
1419 }
|
|
|
1420
|
|
78
|
1421 if (ngx_array_init(&mcf->tables, cf->pool, 1,
|
|
50
|
1422 sizeof(ngx_http_charset_tables_t)) == NGX_ERROR)
|
|
|
1423 {
|
|
|
1424 return NGX_CONF_ERROR;
|
|
|
1425 }
|
|
|
1426
|
|
78
|
1427 if (ngx_array_init(&mcf->recodes, cf->pool, 2,
|
|
|
1428 sizeof(ngx_http_charset_recode_t)) == NGX_ERROR)
|
|
|
1429 {
|
|
|
1430 return NGX_CONF_ERROR;
|
|
|
1431 }
|
|
|
1432
|
|
50
|
1433 return mcf;
|
|
|
1434 }
|
|
|
1435
|
|
|
1436
|
|
|
1437 static void *
|
|
|
1438 ngx_http_charset_create_loc_conf(ngx_conf_t *cf)
|
|
|
1439 {
|
|
|
1440 ngx_http_charset_loc_conf_t *lcf;
|
|
|
1441
|
|
|
1442 lcf = ngx_pcalloc(cf->pool, sizeof(ngx_http_charset_loc_conf_t));
|
|
|
1443 if (lcf == NULL) {
|
|
|
1444 return NGX_CONF_ERROR;
|
|
|
1445 }
|
|
|
1446
|
|
78
|
1447 lcf->charset = NGX_CONF_UNSET;
|
|
50
|
1448 lcf->source_charset = NGX_CONF_UNSET;
|
|
184
|
1449 lcf->override_charset = NGX_CONF_UNSET;
|
|
50
|
1450
|
|
|
1451 return lcf;
|
|
|
1452 }
|
|
|
1453
|
|
|
1454
|
|
|
1455 static char *
|
|
|
1456 ngx_http_charset_merge_loc_conf(ngx_conf_t *cf, void *parent, void *child)
|
|
|
1457 {
|
|
|
1458 ngx_http_charset_loc_conf_t *prev = parent;
|
|
|
1459 ngx_http_charset_loc_conf_t *conf = child;
|
|
|
1460
|
|
78
|
1461 ngx_uint_t i;
|
|
|
1462 ngx_http_charset_recode_t *recode;
|
|
|
1463 ngx_http_charset_main_conf_t *mcf;
|
|
50
|
1464
|
|
184
|
1465 ngx_conf_merge_value(conf->override_charset, prev->override_charset, 0);
|
|
78
|
1466 ngx_conf_merge_value(conf->charset, prev->charset, NGX_HTTP_NO_CHARSET);
|
|
50
|
1467
|
|
|
1468 if (conf->source_charset == NGX_CONF_UNSET) {
|
|
|
1469 conf->source_charset = prev->source_charset;
|
|
|
1470 }
|
|
|
1471
|
|
78
|
1472 if (conf->charset == NGX_HTTP_NO_CHARSET
|
|
|
1473 || conf->source_charset == NGX_CONF_UNSET
|
|
|
1474 || conf->charset == conf->source_charset)
|
|
50
|
1475 {
|
|
78
|
1476 return NGX_CONF_OK;
|
|
50
|
1477 }
|
|
|
1478
|
|
342
|
1479 if (conf->source_charset >= NGX_HTTP_CHARSET_VAR
|
|
|
1480 || conf->charset >= NGX_HTTP_CHARSET_VAR)
|
|
|
1481 {
|
|
|
1482 return NGX_CONF_OK;
|
|
|
1483 }
|
|
|
1484
|
|
78
|
1485 mcf = ngx_http_conf_get_module_main_conf(cf,
|
|
|
1486 ngx_http_charset_filter_module);
|
|
|
1487 recode = mcf->recodes.elts;
|
|
|
1488 for (i = 0; i < mcf->recodes.nelts; i++) {
|
|
|
1489 if (conf->source_charset == recode[i].src
|
|
|
1490 && conf->charset == recode[i].dst)
|
|
|
1491 {
|
|
|
1492 return NGX_CONF_OK;
|
|
|
1493 }
|
|
50
|
1494 }
|
|
|
1495
|
|
78
|
1496 recode = ngx_array_push(&mcf->recodes);
|
|
|
1497 if (recode == NULL) {
|
|
50
|
1498 return NGX_CONF_ERROR;
|
|
|
1499 }
|
|
|
1500
|
|
78
|
1501 recode->src = conf->source_charset;
|
|
|
1502 recode->dst = conf->charset;
|
|
|
1503
|
|
50
|
1504 return NGX_CONF_OK;
|
|
|
1505 }
|
|
78
|
1506
|
|
|
1507
|
|
|
1508 static ngx_int_t
|
|
|
1509 ngx_http_charset_postconfiguration(ngx_conf_t *cf)
|
|
|
1510 {
|
|
184
|
1511 u_char **src, **dst;
|
|
78
|
1512 ngx_int_t c;
|
|
|
1513 ngx_uint_t i, t;
|
|
|
1514 ngx_http_charset_t *charset;
|
|
|
1515 ngx_http_charset_recode_t *recode;
|
|
|
1516 ngx_http_charset_tables_t *tables;
|
|
|
1517 ngx_http_charset_main_conf_t *mcf;
|
|
|
1518
|
|
|
1519 mcf = ngx_http_conf_get_module_main_conf(cf,
|
|
|
1520 ngx_http_charset_filter_module);
|
|
|
1521
|
|
|
1522 recode = mcf->recodes.elts;
|
|
|
1523 tables = mcf->tables.elts;
|
|
|
1524 charset = mcf->charsets.elts;
|
|
|
1525
|
|
|
1526 for (i = 0; i < mcf->recodes.nelts; i++) {
|
|
|
1527
|
|
|
1528 c = recode[i].src;
|
|
|
1529
|
|
|
1530 for (t = 0; t < mcf->tables.nelts; t++) {
|
|
|
1531
|
|
|
1532 if (c == tables[t].src && recode[i].dst == tables[t].dst) {
|
|
|
1533 goto next;
|
|
|
1534 }
|
|
|
1535
|
|
|
1536 if (c == tables[t].dst && recode[i].dst == tables[t].src) {
|
|
|
1537 goto next;
|
|
|
1538 }
|
|
|
1539 }
|
|
|
1540
|
|
|
1541 ngx_log_error(NGX_LOG_EMERG, cf->log, 0,
|
|
342
|
1542 "no \"charset_map\" between the charsets \"%V\" and \"%V\"",
|
|
|
1543 &charset[c].name, &charset[recode[i].dst].name);
|
|
78
|
1544 return NGX_ERROR;
|
|
|
1545
|
|
|
1546 next:
|
|
|
1547 continue;
|
|
|
1548 }
|
|
|
1549
|
|
184
|
1550
|
|
|
1551 for (t = 0; t < mcf->tables.nelts; t++) {
|
|
|
1552
|
|
|
1553 src = charset[tables[t].src].tables;
|
|
|
1554
|
|
|
1555 if (src == NULL) {
|
|
|
1556 src = ngx_pcalloc(cf->pool, sizeof(u_char *) * mcf->charsets.nelts);
|
|
|
1557 if (src == NULL) {
|
|
|
1558 return NGX_ERROR;
|
|
|
1559 }
|
|
|
1560
|
|
|
1561 charset[tables[t].src].tables = src;
|
|
|
1562 }
|
|
|
1563
|
|
|
1564 dst = charset[tables[t].dst].tables;
|
|
|
1565
|
|
|
1566 if (dst == NULL) {
|
|
|
1567 dst = ngx_pcalloc(cf->pool, sizeof(u_char *) * mcf->charsets.nelts);
|
|
|
1568 if (dst == NULL) {
|
|
|
1569 return NGX_ERROR;
|
|
|
1570 }
|
|
|
1571
|
|
|
1572 charset[tables[t].dst].tables = dst;
|
|
|
1573 }
|
|
|
1574
|
|
|
1575 src[tables[t].dst] = tables[t].src2dst;
|
|
|
1576 dst[tables[t].src] = tables[t].dst2src;
|
|
|
1577 }
|
|
|
1578
|
|
230
|
1579 ngx_http_next_header_filter = ngx_http_top_header_filter;
|
|
|
1580 ngx_http_top_header_filter = ngx_http_charset_header_filter;
|
|
|
1581
|
|
|
1582 ngx_http_next_body_filter = ngx_http_top_body_filter;
|
|
|
1583 ngx_http_top_body_filter = ngx_http_charset_body_filter;
|
|
|
1584
|
|
78
|
1585 return NGX_OK;
|
|
|
1586 }
|
